Sumit6307 opened a new issue, #18530:
URL: https://github.com/apache/nuttx/issues/18530
### Is your feature request related to a problem? Please describe.
Yes. The newer flash file system in NuttX, `mnemofs`, is a great
architectural addition but is currently marked as experimental and contains
several structural inefficiencies and incomplete POSIX compliance features. As
noted in multiple `TODO` and `FUTURE TODO` comments across `fs/mnemofs/`, the
current implementation suffers from:
1. **Inefficient Directory Traversal:** `mnemofs_fsobj.c` currently performs
double traversals for path resolution, creating overhead.
2. **Missing Timestamps:** POSIX `mtime`, `atime`, and `ctime` are not fully
implemented/updated during journal atomic commits (noted in `mnemofs.c` and
`mnemofs_journal.c`).
3. **Memory Leaks during Block Flush:** The `flush` operation has known
issues ("does not work properly, and causes memory [leaks]"), which is critical
for flash wear-leveling and system stability in long-running embedded IoT
applications.
### Describe the solution you'd like
I propose a comprehensive stabilization pass over `mnemofs` focusing on
robustness, POSIX standard compliance, and maintaining strict
self-compatibility. The planned solution includes:
* **Single Traversal Refactoring:** Rewriting the directory lookup
mechanisms to utilize a single-pass traversal, decoupling the LRU cache
dependencies from Journal atomic commits to increase lookup throughput without
increasing RAM footprint.
* **Timestamping Integration:** Properly passing POSIX time references down
through the VFS inode layer into the `mnemofs` master nodes, updating
timestamps gracefully without causing massive flash rewrite amplification.
* **Flush Mechanism Overhaul:** Investigating and rewriting the block flush
mechanism to permanently resolve memory leaks, verifying stability under heavy
multi-threaded write loads over long periods.
* **Hardware Validation:** Adding extensive runtime test coverage using
`apps/ostest` and `apps/fs/fstests` verified on real hardware, generating
complete build and runtime logs to prove zero regressions.
This work will be done iteratively through small, highly-focused,
non-breaking Pull Requests to allow manageable reviews by the maintainers.
### Describe alternatives you've considered
I propose a comprehensive stabilization pass over `mnemofs` focusing on
robustness, POSIX standard compliance, and maintaining strict
self-compatibility. The planned solution includes:
* **Single Traversal Refactoring:** Rewriting the directory lookup
mechanisms to utilize a single-pass traversal, decoupling the LRU cache
dependencies from Journal atomic commits to increase lookup throughput without
increasing RAM footprint.
* **Timestamping Integration:** Properly passing POSIX time references down
through the VFS inode layer into the `mnemofs` master nodes, updating
timestamps gracefully without causing massive flash rewrite amplification.
* **Flush Mechanism Overhaul:** Investigating and rewriting the block flush
mechanism to permanently resolve memory leaks, verifying stability under heavy
multi-threaded write loads over long periods.
* **Hardware Validation:** Adding extensive runtime test coverage using
`apps/ostest` and `apps/fs/fstests` verified on real hardware, generating
complete build and runtime logs to prove zero regressions.
This work will be done iteratively through small, highly-focused,
non-breaking Pull Requests to allow manageable reviews by the maintainers.
### Verification
- [x] I have verified before submitting the report.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
